Does a Growth Mindset Really Work? What the Research Found

In the late 1990s, Claudia Mueller and Carol Dweck ran a series of experiments with fifth-graders that would go on to shape a generation of classroom posters. Children solved a set of problems and then heard a single sentence of praise. Some were told they must be smart. Others were told they must have worked hard.

Then the children met harder problems, and failed. Those praised for intelligence afterwards enjoyed the task less, persisted less and performed worse than those praised for effort. They were also more likely to describe intelligence as something fixed. Mueller and Dweck published six studies along these lines in 1998.

Dweck turned this line of work into a popular book in 2006, and the idea spread fast: people who believe abilities can grow - a growth mindset - learn more than people who believe abilities are set in stone - a fixed mindset. Schools around the world adopted it. Then researchers started checking how big the effect really was.

The first big check: two meta-analyses

In 2018, Sisk, Macnamara and colleagues pooled the evidence in two meta-analyses published in Psychological Science. The first asked whether students' mindsets are related to their achievement at all: 273 studies, more than 365,000 students. The second asked whether programmes designed to teach a growth mindset actually raise achievement: 43 studies, about 57,000 students.

Their conclusion was blunt: overall effects were weak in both. The link between believing ability can grow and doing well at school exists, but it is small, and teaching the belief did little for the average student. There was one interesting exception. Students from low socioeconomic backgrounds, and students who were already academically at risk, appeared to benefit more. That is roughly what the theory predicts: a belief about improving should matter most when you are struggling.

The national experiment

The most careful test came a year later. David Yeager, Dweck and a large team ran the National Study of Learning Mindsets, published in Nature in 2019. They drew a random sample of 65 regular public schools across the United States, and 12,490 ninth-graders were individually assigned at random either to a growth mindset programme or to a control programme.

The intervention was tiny: two self-administered online sessions of about 25 minutes each, roughly three weeks apart. Independent companies collected and processed the data, the analyses were registered in advance, and a blinded statistical analysis was used to corroborate the results. These are exactly the safeguards that much of the earlier literature lacked.

The results were real, but modest and specific:

• Among lower-achieving students, grades in core subjects rose by about 0.10 grade points, on the American scale where an A is worth 4. The authors put this at a standardised effect of 0.11.
• Among higher-achieving students there was no measurable effect on grades.
• In the schools with course data, the share of students taking advanced mathematics the following year rose from 33% to 36%.
• Effects on grades were largest in schools where peer norms supported taking on challenges. Where the surrounding culture did not, the message did not stick.

This is what a small effect looks like in practice. A tenth of a grade point will not transform one student's life. Spread across a whole country for the price of under an hour, it is not nothing either.

The sceptical view

Not everybody reads the evidence the same way. In a 2023 review in Psychological Bulletin, Macnamara and Burgoyne examined 63 studies with nearly 98,000 students and checked them against best practices for drawing causal conclusions. They found major shortcomings in design and reporting, and signs of publication bias. The average effect was small (d = 0.05) and stopped being statistically significant once they corrected for possible bias. In the six highest-quality studies it was 0.02, indistinguishable from zero. They also found that authors with a financial incentive to report positive results published larger effects.

Tipton, Yeager and colleagues replied in the same journal that averaging across very different schools and programmes is the wrong question: meta-analyses should examine heterogeneity, meaning where and for whom an intervention works. The debate is still open. But both sides agree on one thing: the grand claims of the early popular coverage were not supported. A growth mindset is not a switch that turns a struggling student into a star.

What this does and does not mean

It does not mean that beliefs about ability are irrelevant. The correlation is small but it is there, and there are signs that believing you can improve matters more when things are hard.

It does not mean that praising effort is magic. Praising effort that did not produce learning, or telling people to simply believe in themselves, is not what the research tested. The programmes taught a specific idea: that abilities can be developed with practice and good strategies, and that struggle is part of learning.

It does mean that context counts. The same message worked in some schools and not in others, depending on whether the surrounding culture rewarded challenge-seeking. A mindset is not only inside your head.

And it means that the honest size of the effect is small. Most of what predicts achievement - what you already know, the quality of teaching, time, support - lies elsewhere.

What this means for you

If you take our growth mindset test, treat your score as a description of how you currently think about ability, not as a forecast of your success. A few ideas grounded in the evidence:

• Look at what you do after a setback, not at the slogan you agree with. The studies measured behaviour: persisting, choosing harder courses.
• Pair belief with strategy. Believing you can improve helps most when you also know how: practice, feedback, a better method. Our achievement goals test looks at whether you tend to aim at learning or at looking competent, a closely related question.
• Notice your environment. If the people around you treat effort as embarrassing, the evidence suggests that matters. Spending time where challenge is normal may do more than any pep talk.
• Keep the ideas apart. Believing that ability can grow is not the same as feeling capable of a specific task. That second belief is what our self-efficacy test measures, and it deserves its own attention.

A small, real effect is still worth having. It just is not the revolution the posters promised.